| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Oak Ridge Conference |
1 NAXCOR, Inc., 320 Logue Ave., Suite 200, Mountain View, CA 94043.
aAuthor for correspondence. Fax 650-641-1918; e-mail risa{at}naxcor.com
Abstract
Background: Current technologies for assessing genetic deletions and duplications of greater than one kilobase are labor-intensive or rely on PCR-based methods, and none offers the ability to simultaneously detect dosage abnormalities, assess 5'-to-3' cytosine-guanosine (CpG) methylation, and interrogate single-nucleotide polymorphisms (SNPs). We describe a high-throughput platform for direct gene-dosage determination capable of concurrent assessment of other forms of gene modification.
Methods: We used a light-activated interstrand nucleic acid cross-linking system (XLntTM technology) to determine gene dosage at the 15q11-q13 deletion/duplication locus. We incorporated restriction enzyme digestion of genomic DNA into the method to assess CpG methylation in parallel with gene dosage. For method validation we used DNA from 31 cell lines with previously characterized 15q11-q13 gene dosage and parental origin status. Diagnostic cutoffs were set at 0.5 ± 0.15, 1 ± 0.15–0.25, and 2 ± 0.3.
Results: Dosage-only experiments showed discrimination of deletions (n = 21) from healthy controls (NCs; n = 27) in all samples. Five of 49 samples gave results outside of specification. Concurrent evaluation of dosage and CpG methylation yielded dosage results within specification for 18 of 19 deletion and 8 of 12 NC samples. Paternal deletion and NC methylation pattern results were within specification in 17 of 19 and 9 of 12 runs, respectively. No overlap was demonstrated between value sets for the two groups.
Conclusions: The XLnt technology provides a rapid, high-throughput platform for the accurate determination of gene dosage. The flexibility of this technology allows parallel interrogation of gene dosage, CpG methylation, and SNPs.
Current methods for genetic quantification capable of discriminating among one, two, and more gene copies in mammalian cells are generally expensive, tedious, and require the services of highly skilled personnel for their performance and interpretation. The current gold standard in clinical diagnostics for the evaluation of chromosomal submicroscopic deletions and duplications is fluorescence in situ hybridization. Newer methods promising faster turnaround include comparative genomic hybridization against metaphase chromosome spreads (1) or arrayed genomic clones (2)(3)(4), and quantitative PCR using real-time analysis of logarithmic-phase amplicon accumulation to determine relative target amounts (5)(6).
Evaluation of epigenetic DNA modifications is an emerging area of molecular diagnostics. Methylation abnormalities of sequence-specific 5'-to-3' cytosine-guanosine (CpG) 1 dinucleotides are associated with inherited developmental disorders and are particularly important in the sequential accumulation of somatic mutations in the development of cancer (7)(8)(9). Methods for determining CpG methylation status include Southern blotting using methylation-specific restriction enzymes and, recently, PCR after bisulfite modification of unmethylated cytosine bases (10). Comprehensive molecular evaluation of germline or somatic mutations and polymorphisms will require the ability to assess multiple mechanisms of mutations in parallel. No technology currently offers the simplicity and flexibility necessary for high-throughput, concurrent determination of simple nucleotide alterations, whole-gene and larger deletions or duplications, and CpG methylation.
The deletion/duplication locus at 15q11-q13 serves as a model system for developing techniques for the concurrent assessment of gene dosage and CpG methylation status (11)(12). This region is prone to spontaneous deletion and duplication events of characteristic size attributable to the presence of flanking low-copy repetitive regions (13). In addition, the entire region is subject to gametic imprinting such that particular genes within the critical interval are expressed exclusively from the chromosome contributed by a specific parent. The Prader–Willi syndrome (PWS) is characterized by moderate to severe mental retardation, neonatal hypotonia, profound obesity, and dysmorphic features, whereas the features of Angelman syndrome (AS) comprise severe mental retardation with autistic features and seizure disorder with normal body habitus. In both syndromes, 
70% of individuals have deletions of the identical genetic material from chromosome 15q11-q13; in the former instance, deletions are found on the paternally derived chromosomal region, whereas in the latter, the deleted material is from the homolog inherited from the mother. Duplication events associated with this region produce a subtler phenotype, including a form of autism associated with duplications occurring predominantly on the maternal chromosome (14)(15).
Confounding the molecular diagnosis of PWS and AS is the observation that nearly identical phenotypes are produced by other mechanisms. Both phenotypes have been observed in the setting of chromosome 15 uniparental disomy, in which an offspring inherits both chromosome 15 homologs from one parent and none from the other. Whereas paternal disomy is rarely seen in AS, maternal disomy accounts for 28% of the cases of PWS. A third cause of both syndromes is small deletions in the "imprinting center" conferring the epigenotype, or methylation pattern, of the alternate parent on the chromosome 15 homolog (16)(17). Molecular testing for PWS and AS has necessarily proceeded in a stepwise fashion to investigate both deletions and epigenotype (18)(19)(20). For efficient and accurate molecular diagnosis of PWS/AS and duplication 15q11-q13 syndromes, there exists a clinical need for a rapid, cost-effective technology that will allow for the parallel ascertainment of gene dosage and methylation status.
We have adapted the nucleic acid cross-linking (XLntTM) technology, a photo-cross-linking oligonucleotide hybridization methodology (developed by NAXCOR) previously established for single-nucleotide polymorphism (SNP) detection, to this application (21)(22). The XLnt technology comprises incorporation of a photoactivatable coumarin-based nucleotide analog in place of an adenosine residue into an oligonucleotide complementary to a target DNA region. Within the context of an oligonucleotide-nucleic acid hybridization complex, exposure to ultraviolet light causes a covalent bond to form between the XLnt moiety and the opposing thymidine residue. The covalently bound probe-target complex is rendered capable of withstanding high-stringency wash conditions, producing a marked increase in the signal-to-background ratio and, thus, greater sensitivity. The factor V Leiden (21) and hereditary hemochromatosis HFE C282Y and H63D (22) SNP assays use separate fluoresceinlabeled reporter probes for signal amplification and a single biotinylated capture probe that confers allele specificity; both oligonucleotide probe types contain the cross-linking molecule. Signal amplification is accomplished by sequential incubation with an anti-fluorescein antibody-alkaline phosphatase conjugate and the alkaline phosphatase substrate, AttoPhos®. The assay design and signal generation methodology modified for this study are shown in Fig. 1
.
|
Materials and Methods
oligonucleotide synthesis
A 1980-bp unique sequence from within the duplication/deletion interval including small nuclear ribonucleoprotein polypeptide N gene (SNRPN) exon 1, known to be reliably differentially methylated between the maternally and paternally derived chromosomes 15, was identified (23). Two discrete probe sets were designed from this region. Both sets contained an overlapping panel of four to six fluorescein-labeled reporter probes complementary to sequences downstream from a group of three sites recognized by the methylation-sensitive restriction enzyme HpaII. The first probe set contained a pair of biotin-labeled capture probes that recognized sequences upstream of these restriction sites, whereas the second set contained capture probes complementary to sequences downstream of these sites and on the same side as the panel of reporter probes. A third probe set comprising two capture probes and four reporter probes was designed from a control locus that contained intronic sequences from the ankyrin 2 gene (ANK2) at 4q25 and excluded HpaII sites.
The overall probe design is illustrated in Fig. 2
. The nucleotide numbers for the probes correspond to sequences from either the SNRPN gene or the B240N9 clone from 4q25, for which the GenBank accession numbers are U41384 and ACC004057, respectively. Probe sequences were as follows: 5'-SNRPN capture probes, nucleotides (nt) 15290–15335 and 15191–15223; 3'-SNRPN capture probes, nt 16018–16047 and 16058–16095; reporter probes common to the 3'- and 5'-SNRPN probe sets, nt 15668–15694, 15763–15782, 15815–15839, and 15606–15628; additional 5'-SNRPN reporter probes, nt 15579–15603 and 15989–16015; 4q25 control capture probes, nt 22741–22779 and 22311–22348; 4q25 control reporter probes, nt 22361–22382, 22637–22658, 22700–22723, and 22834–22860. Capture probes were biotinylated at the 3' end. Reporter probe synthesis was modified from the previous method (21) by introducing 40 amino-spacer groups to the 3' tail of the oligonucleotide backbone during probe synthesis. The probes were deprotected and purified as described (21). Fluorescein groups were incorporated into the reporter probes by incubating the polyamine probes with the amino-reactive reagent fluorescein-NHS ester (FLUOS; Roche Applied Science). Excess FLUOS was removed through repeated rounds of centrifugal concentration. The extent of fluorescein incorporation was determined spectrophotometrically to be, on average, 20–25 molecules of fluorescein per probe. Two coumarin-based cross-linking nucleotides were incorporated into each probe in place of an adenosine nucleotide near the 3' and 5' ends as described (21).
|
human subjects
DNA was obtained from lymphoblastoid cell lines derived from individuals with previously characterized abnormalities of 15q11-q13 {12 PWS patients with deletions of 15q11-q13 and two patients with isodicentric chromosomes 15 [karyotype 47,XX,+dic(15)(pter>q13::q13>pter)]} and 16 healthy controls (NCs) from the Coriell Cell Repository maintained by the National Institute of General Medical Sciences. Two additional lymphoblastoid cell lines created from NC individuals were provided by Dr. N.C. Schanen (UCLA, Los Angeles, CA). Informed consent for the creation and distribution of these cell lines was obtained under an Institutional Review Board-approved protocol (UCLA).
hybridization assay
The hybridization assay was performed through an adaptation of previously described procedures (21)(22). Briefly, lymphoblasts were isolated by centrifugation from cell cultures. Cell counts were obtained on all lymphoblast specimens and ranged from 0.67 x 106 to 15.6 x 106 cells/well. For parallel dosage and methylation assays, genomic DNA was extracted from lymphoblasts (Puregene; Gentra Systems). Genomic DNA (210–425 µg) was digested overnight with the restriction enzyme HpaII (1 U/µg of DNA; New England Biolabs), precipitated with ethanol, and resuspended in leukocyte lysis buffer (0.28 mol/L NaOH); DNA aliquots ranged from 32 to 65 µg/well. For dosage-only assays, lymphoblast pellets were suspended directly in leukocyte lysis buffer and used immediately or stored frozen at -70 °C before use.
Alkalinized samples were boiled for 20 min immediately before assay performance to fragment the genomic DNA. Processed samples were analyzed in duplicate in four (dosage-only analysis) or six (parallel dosage/methylation analysis) wells each of a 96-well polypropylene microtiter plate. Each assay plate also contained two negative controls and two positive controls per each probe set. For dosage-only studies, up to 10 samples could be run simultaneously; for dosage/methylation studies, a total of 6 samples could be analyzed per plate. Probe solutions were prepared as described (21). Final concentrations used were 0.5 pmol/well for each capture probe and 0.2 pmol/well for each reporter, with the exception of aliquots for the negative controls for the dosage/methylation experiments, for which the capture probes were omitted. Aliquots of each probe solution were added in duplicate to each sample well as well as to the negative and positive control wells.
After neutralization of the solutions, photo-cross-linking, and addition of the streptavidin-coated magnetic beads (Dynal), the beads were washed first with a prewash [1 g/L sodium dodecyl sulfate, 0.1x standard saline citrate (SSC; 150 mmol/L sodium chloride–15 mmol/L trisodium citrate), 0.01 mL/L Tween® 20], then with the gene-dosage high-stringency wash (500 mL/L formamide, 5 mL/L Tween 20, 0.1x SSC), and finally with a preantibody wash (1x SSC, 1 mL/L Tween 20). The washed beads were incubated in the presence of anti-fluorescein antibody-alkaline phosphatase conjugate (Dako), washed four times, and finally resuspended in a solution containing AttoPhos (Promega). The fluorescence signal attributable to the hydrolysis of AttoPhos was determined by use of a FluoroCountTM microplate reader (Packard Instrument Company).
data analysis
Two separate experimental designs were used. For dosage determination alone at 15q11-q13, leukocyte pellets were assayed directly without prior DNA extraction. Samples were assayed separately with the 3'-SNRPN and 4q25 probe sets. Negative controls in these experiments comprised lysis buffer alone. Leukocyte pellets from phenotypically normal individuals were included in each plate as positive controls. Overall SNRPN gene dosage was determined by calculating the ratio of net sample signal (NSS; mean sample signal with mean background signal subtracted) from the 3'-SNRPN assay to that from the 4q25 assay for each sample. The ratios were normalized by dividing by 1.2, which reflected the difference in signal intensity between the two probe sets and was derived from assaying a set of NCs. These normalized values represented the normalized signal ratios (NSRs). Cell lines were assayed either once or twice, depending on the rapidity of cell growth. The results presented represent data collated from all runs, except from two samples that consisted of fewer than 1 x106 cells/well, which were omitted from the overall analysis.
For concurrent dosage/methylation determination, HpaII-digested DNA was assayed with each of the three described probe sets. Positive controls comprising undigested DNA from a phenotypically normal individual were included in each experiment for normalization. Negative controls for these experiments consisted of undigested human or calf thymus DNA (comparable values were obtained for both) assayed with probe sets excluding capture probes. Control samples were prepared and processed identically as with the samples.
For each sample and positive control, the NSS was determined for each of the three probe sets. The NSS ratios of 3'-SNRPN to 4q25 and 5'-SNRPN to 3'-SNRPN were obtained for each sample, and each of these values was normalized against the ratios obtained for the positive control sample to give the NSR. The 3'-SNRPN-to-4q25 NSR determined the overall gene dosage for the 15q11-q13 region, whereas the 5'-SNRPN-to-3'-SNRPN NSR reflected the fraction of SNRPN gene exon 1 methylation. Taken together, the calculated ratios provided a comprehensive profile of the 15q11-q13 region. The expected relative values and the genotypes and phenotypes to which they correspond are summarized in Table 1
.
|
assay precision study
Inter- and intraassay imprecision was assessed for both the dosage-only method performed using lysed cell pellets as template and for the combined methylation-dosage experiments using digested DNA; material for both studies was drawn from NCs. For the dosage-only experiments, suspended cells were aliquoted into separate tubes and centrifuged, and aliquots were suspended in lysis buffer. For the methylation-dosage experiments, DNA was digested and aliquoted before DNA precipitation. For each study, five aliquots were assayed in separate plates in separate runs to assess between-run precision. To assess within-run precision, five aliquots of each template type were assayed in a single plate in one run.
Results and Discussion
The utility of a hybridization-based gene-dosage assay relies on its ability to determine accurately the relative target amounts of both test and control loci. This goal is achieved most reliably by use of a hybridization-dependent signal-generation scheme in which signal intensity correlates linearly with locus copy number. Assay reliability is promoted under conditions exhibiting consistently low background signal and high signal-to-noise ratios. Typically, oligonucleotide hybridization assays must use low-stringency wash conditions to maintain the integrity of the hydrogen-bonded oligonucleotide-target complex. Consequently, nonspecific hybridization complexes that survive the washing steps will confound the accurate quantification of specific hybrid complexes. By covalently linking the probe to the target using the XLnt photo-cross-linking technology, the initially noncovalent hybridization complex is transformed into a covalent molecule in which the association between target and probe is stable to high-stringency wash conditions. Application of this technology coupled to a standard linear signal amplification method afforded by alkaline phosphatase allows accurate quantification of gene dosage.
The gene-dosage and methylation assays described herein are, in essence, both "dosage assays" in that both determinations rest on detecting fluorescent signals produced by hapten-modified reporter probes, where the signals indicate the relative amounts of gene target and specifically methylated CpG sites, respectively. That is, the dosage ratio reflects the number of SNRPN sequences relative to the diploid control locus at 4q25, whereas the methylation ratio reflects the number of SNRPN loci specifically methylated at HpaII sites within exon 1 relative to the totality of SNRPN sequences. The experimental design and signal amplification methodology of the photo-cross-linking assay are illustrated in Fig. 1
.
Results from dosage-only experiments using cell samples from 12 individuals harboring deletions at 15q11-q13, 18 NCs, and 1 individual with a twofold duplication of 15q11-q13 are presented in Table 2
. The predicted NSR value for deleted samples is 0.5, whereas that for samples possessing a normal diploid copy number is 1. Both deletion and control experimental value sets were observed to approximate gaussian distributions with calculated 2 SD values of 0.15 and 0.18, respectively. On the basis of these results, the diagnostic ranges for this study were fixed at the predicted value ± 0.15. Eighteen of 21 (85.7%) deleted samples and 25 of 27 (92.6%) NCs fell within the diagnostic tolerance. The ranges for the deletion samples and NCs were discrete, with no overlap seen between out-of-specification values (Fig. 3
). The ability of the current method to assess gene copy number accurately is, therefore, clearly demonstrated.
|
|
Reliable dosage determinations were obtained on aliquots containing at least 1.1 x 106 cells/well, which translates to a minimum required blood volume for four sample wells of 1 mL for individuals with leukocyte counts within the reference limits. However, because leukocyte recovery from a red cell lysis is, in practice, incomplete, we estimate that the current assay will more likely require a minimum whole blood volume of 1.5 mL.
The results for the concurrent determination of gene dosage and CpG methylation using DNA derived from 12 PWS individuals, 9 NCs, and the 2 individuals with marker 15 chromosomes are summarized in Tables 3
and 4
. Diagnostic tolerances were set slightly less stringently for the high-end expected NSR values, as follows: haploid range, expected value of 0.5 ± 0.15; diploid range, expected value of 1.0 ± 0.20; tetraploid range, expected value of 2.0 ± 0.3. Given these cutoffs, 18 of 19 (94.7%) deletion samples and 8 of 12 (66.7%) NCs fell within these dosage ratio ranges. Two dosage values for cell lines harboring a fourfold gene dosage at 15q11-q13 have also been included (Fig. 4
).
|
|
|
The diagnostic range for the NC methylation ratios were set as for the deletion sample dosage range: 0.35–0.65, surrounding the theoretical value of 0.5. The methylation ratio diagnostic range for the deletion samples was set less stringently, at 0.75–1.25. In the case of the deleted samples, only a single chromosome was subject to evaluation; therefore, it was necessary only to discriminate complete methylation (expected methylation ratio of 1) from complete absence of methylation (expected methylation ratio of 0). For normal copy dosage, however, it was necessary to discriminate hemimethylation (normal biparental inheritance; expected methylation ratio of 0.5) from both complete methylation (maternal disomy; expected methylation ratio of 1) and absence of methylation (paternal disomy; expected methylation ratio of 0). Consequently, the more stringent diagnostic cutoffs were preserved for the normal-dosage samples. Results of the methylation experiments, excluding the duplication samples for which parental origin was not previously determined, are shown in Fig. 5
. Of 31 samples tested, 17 of 19 (84.2%) paternal deletions and 9 of 12 (75%) NCs gave values falling within the diagnostic ranges. Results were excluded from four NCs with macroscopic turbidity and one PWS individual. The latter sample gave results consistent with normal 15q11-q13 dosage, whereas repeat values were within expected ranges.
|
The inter- and intraassay imprecision was studied with cell lysates and DNA pooled from control individuals and aliquoted into equivalent-sized samples. For within-run precision assessment, five samples were assayed in one plate, whereas between-run precision was assessed by assaying each of five samples individually in five separate plates. CVs determined from the 3'-SNRPN-to-4q25 NSS ratio (dosage NSS ratio) and the 5'-SNRPN-to-3'-SNRPN NSS ratio (methylation NSS ratio) are shown in Table 5
. The sample size for this study was small, but the within-run CVs of 12–13% should reflect a realistic estimate of the imprecision of these assays.
|
The combined methylation and dosage experiments were conducted using a minimum of 40 µg of DNA/well, with six wells required for assaying the sample with each of the three probe sets in duplicate. This requirement should translate to a starting whole blood volume of, on average, 5–10 mL. Ongoing work with improved methods of sample preparation and probe design is expected to improve sensitivity, thereby decreasing the requisite sample volume.
The XLnt photo-cross-linking technology has previously been applied to the detection of the factor V Leiden and HFE C282Y and H63D SNPs (21)(22). Differential hybridization of allele-specific capture probes and subsequent cross-linking before the washing steps allows SNP detection assays to be performed under conditions identical to those of the dosage assays described above. Therefore, this methodology should allow concurrent detection of SNPs, gene dosage, and CpG methylation in a single platform. XLnt photo-cross-linking assays should also be amenable to performance in an automated format.
The XLnt photo-cross-linking technology is ideally suited for providing rapid, accurate, automated, and non-PCR-based determination of germline dosage abnormalities. It offers obvious advantages over fluorescence in situ hybridization and quantitative PCR with respect to speed and complexity. Importantly, the flexibility of the technology allows the concurrent determination of multiple mechanisms of mutation in a single platform, including both SNPs and CpG methylation status.
Footnotes
1 Nonstandard abbreviations: CpG, 5'-to-3' cytosine-guanosine DNA sequence; PWS, Prader–Willi syndrome; AS, Angelman syndrome; XLnt, cross-linking nucleotide; SNP, single-nucleotide polymorphism; SNRPN, small nuclear ribonucleoprotein polypeptide N gene; nt, nucleotide(s); NC, healthy control; SSC, standard saline citrate; NSS, net sample signal derived from the mean sample signal corrected for background; and NSR, normalized sample ratio derived from the ratio of NSS from two probe sets normalized to control values. 
References
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  C. French, C. Li, C. Strom, W. Sun, R. Van Atta, B. Gonzalez, and M. Wood Detection of the Factor V Leiden Mutation by a Modified Photo-Cross-Linking Oligonucleotide Hybridization Assay Clin. Chem., February 1, 2004; 50(2): 296 - 305. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
) and NCs (
).
), and NCs (